Character Input Using the Motion Sensor on a Smartwatch.

The screen of a smartwatch is small, which creates operational issues such as erroneous input and screen occupancy of the keyboard. Therefore, an input method, that does not depend on screen tapping but instead utilizes taps outside the touch screen and a wrist operation, is proposed herein. Five types of tapping movements were used in this study, which were aimed outside the screen and involved wrist movements. With respect to the tapping movements outside the screen, the tapped parts were placed in different locations, and tapping was performed toward the center of the screen. The wrist movement comprised a series of movements ranging from pronation to supination. These movements were assumed to be the character input. As five types of tap movements were used, five vowels of flat kana can be input in a single movement. Ten selections can be made with one input, depending on the tapping movements outside the screen and the presence or absence of wrist movement. By doing this twice, 100 selections can be made, and QWERTY array keyboard input is possible. Accelerometers and angular velocity sensors, built into the smartwatch, were used to determine the movements. In the experiment, the subjects performed each movement and data were acquired. As a result of the analysis, the off-screen tap movement was determined by the acceleration in the x-axis and y-axis. It was suggested that it is possible to discriminate whether the wrist movement is based on the angular velocity of the x-axis. A real-time discrimination experiment based on these revealed that the average discrimination rate was 81.7%. The proposed motions can determine several choices, and the efficacy of the proposed method as a viable input method for smart watches was clearly demonstrated.